Competition between two planktonic rotifer species at different temperatures: an experimental test

Transcription

1 Freshwater Biology (2006) 51, doi: /j x Competition between two planktonic rotifer species at different temperatures: an experimental test CLAUS-PETER STELZER Department of Physiological Ecology, Max Planck Institute of Limnology, Plön, Germany SUMMARY 1. The effect of temperature on the outcome of resource competition between two planktonic rotifers (Synchaeta pectinata and Brachionus calyciflorus) was investigated in laboratory experiments. In addition to the competition experiments, several physiological variables and their temperature-dependence were characterised, including ingestion rate and starvation tolerance. 2. Because of a lower threshold food level (TFL) for population growth for the food algae Cryptomonas erosa, Synchaeta was predicted to be the superior competitor at low temperatures (12 C). In contrast, Brachionus had a lower TFL at 20 C and was predicted to be competitively superior at this temperature. 3. In both rotifer species, ingestion rates increased with temperature, but the increase was much more pronounced in Brachionus. Ingestion rates of Brachionus at temperatures from 8 to 24 C were always higher than in Synchaeta (up to 4.6-fold). 4. Starvation resistance reduced with temperature in both rotifer species. At all temperatures investigated (12, 16 and 20 C) Brachionus could survive starvation for longer than Synchaeta. This difference was strongest at 12 C (5.8 days versus 2.5 days). 5. In the first competition experiment, food was supplied at 48 h-intervals. Brachionus displaced Synchaeta at both experimental temperatures (12 and 20 C). Competitive exclusion of Synchaeta at the lower temperature was probably because of large fluctuations in algal densities that resulted from the long intervals between feeding, a condition that favoured Brachionus because of its higher starvation resistance. 6. In the second competition experiment, one third of the food suspension was renewed every 8 h, resulting in a much better approximation to a continuous resource supply. At 12 C Synchaeta and Brachionus coexisted for more than 1 month and the densities of both rotifer species were significantly lower in the presence of their competitor. In contrast to expectations, Brachionus was able to persist even when Cryptomonas concentrations fell below its TFL. This was probably because Brachionus was using detritus and associated bacteria as additional food sources, which were present in the cultures during the later phase of the experiment. 7. Autocorrelation analysis of the temporal changes in egg ratios revealed significant periodic cycles in Synchaeta during the second competition experiment. A possible explanation for this is the fecundity schedule of Synchaeta, in which reproduction is highly concentrated in a few age classes. According to demographic theory, such a life cycle feature can cause slower convergence to a stable age distribution. Keywords: microcosm, population growth, resource competition, Rotifera, temperature change Correspondence: Claus-Peter Stelzer, Institute for Limnology of the Austrian Academy of Sciences, Mondseestrasse 9, A-5310 Mondsee. claus-peter.stelzer@oeaw.ac.at Ó 2006 The Author, Journal compilation Ó 2006 Blackwell Publishing Ltd 2187

2 2188 C.-P. Stelzer Introduction Ecologists have long tried to understand the distribution of species in relation to temperature clines. The importance of understanding such patterns has renewed emphasis because of research on the effects of global warming (Hughes, 2000; Stenseth et al., 2002). One formerly popular approach to predict changes in the composition of species was the climate-mapping approach, which consists of mapping the distribution of a species in climate-space and then, if the position of that climate-space changes, predicting the shift of the species distribution. This approach, as well as other simplistic measures of the temperature preference of a species (e.g. abundances, physiological tolerance, or growth performance in relation to temperature), are however inadequate for predictions at the community level, because biological processes like predation, competition or dispersal can have significant influences on patterns of species abundances (Davies et al., 1998a,b; Jiang & Morin, 2004). In addition, some of these processes are temperature-dependent themselves; for instance, competition coefficients may change along temperature gradients. Thus a deeper understanding of the processes affecting the distribution of species in relation to temperature may require a characterisation of the actual mechanisms underlying such temperature-related changes. Resource competition theory (Tilman, 1982) has been very successful in predicting the outcome of competition using a mechanistic approach. Rather than lumping competitive effects into the competition coefficient, the theory predicts the outcome of competition between two or more species on the basis of individual resource requirements. The simple prediction in the case of n competitors and one resource is that the species with the lowest food requirement to maintain zero population growth (threshold food level, TFL) will always win. Resource competition theory can also be extended to more complex situations, such as where more than one resource type is available. Its successful application has been demonstrated for several organisms, including algae and mesozooplankton (Tilman, 1982; Rothhaupt, 1988; Kreutzer & Lampert, 1999). Threshold food levels are not species-specific constants but can vary depending on several conditions. For example Rothhaupt (1990a) found that the TFLs of two Brachionus species depend on the size of their food algae. The larger Brachionus calyciflorus Pallas showed the lowest TFL when it was cultured with the relatively large algae Chlamydomonas sphaeroides Gerloff, whereas the smaller Brachionus rubens Ehrenberg had the lowest TFL with the relatively small algae Monorhaphidium minutum Nägeli (Rothhaupt, 1990a). Rothhaupt (1990b) suggested that this was because of the differential efficiency of the two rotifers to ingest particles of different sizes and this, in turn, seemed to depend on the relationship between the size of the rotifer and of its food algae. Food algae-dependent TFLs were also found by Ciros-Perez, Carmona & Serra (2001) for three species of the Brachionus plicatilis complex, although in this study no clear relationship with animal size was apparent. Temperature is another environmental factor that can modify a species TFL (Achenbach & Lampert, 1997; Stelzer, 1998; Weisse et al., 2002). In most cases TFLs stay constant for a range of temperatures and increase at higher temperatures. In some cases the TFL may also be elevated at very low temperatures, leading to a U-shaped pattern of TFL in relation to temperature (e.g. Weisse et al., 2002). An application of resource competition theory to species distributions along temperature clines involves measurements of the TLFs of the different species at different temperatures. If the TFL-temperature isoclines of two species run parallel, no shifts in competitive dominance are expected. Such a result was obtained by Achenbach & Lampert (1997) for four differentially sized Daphnia species. However, if such TFL-temperature isoclines do cross, resource competition theory predicts that one species will be superior at low temperatures while the other will be superior at high temperatures. Such a situation was described by Tilman, Mattson & Langer (1981) for two diatom species. The authors found that Asterionella formosa Hassal competitively displaced Synedra ulna Ehrenberg at temperatures below 20 C but that the situation was reversed at temperatures above 20 C. This outcome was predicted by measurements on the silicate growth kinetics of the two algae at different temperatures (Tilman et al., 1981). In the present paper I examine an analogous situation in two heterotrophic consumers. Laboratory experiments and field data suggest that the two rotifers Synchaeta pectinata Ehrenberg and B. calyciflorus have different thermal preferences. In a study involving several hundred freshwater habitats,

3 Methods Temperature and competition of rotifers 2189 Fig. 1 Temperature-dependence of the threshold food levels of Synchaeta pectinata and Brachionus calyciflorus feeding on Cryptomonas erosa. Redrawn from Stelzer (1998). sampled during different years and different seasons, Berzinš & Pejler (1989) established frequency distributions of 225 rotifer species in relation to water temperature. The highest abundances of Synchaeta were found at 12 C, while Brachionus was most abundant at 18 C (Berzinš & Pejler, 1989). Additionally, an experimental study suggested that Synchaeta uses its resources most efficiently at low temperatures: Stelzer (1998) found that the TFL of Synchaeta increased almost 10-fold from 12 to 24 C (see Fig. 1). In contrast, the TFL of Brachionus showed only minor differences over the same temperature range, being lowest at C and slightly elevated at 12 and 24 C. As the isoclines connecting the TFLs at the different temperatures intersected at 15 C, it was predicted that Synchaeta should outcompete Brachionus at temperatures below 15 C while Brachionus should be the superior competitor above 15 C, provided that both species compete for the same food resource (Stelzer, 1998). The aim of this paper was to test these predictions in competition experiments at two temperatures. The hypotheses were: (i) Synchaeta will competitively displace Brachionus at 12 C and (ii) Brachionus will competitively displace Synchaeta at 20 C. In addition to the competition experiments, the temperaturedependence of feeding rates and starvation times were characterised for both rotifer species. This allowed the construction of a competition model to predict the time course of competitive exclusion. Resource concentrations were measured in each replicate to determine whether resource use and population dynamics of the two species were in accordance with theoretical predictions. The S. pectinata clone was derived from an individual stem female that was isolated from Schöhsee (northern Germany). This clone has been characterised in detail in several previous publications and information on various life history parameters is available: population growth and TFLs at different temperatures (Stelzer, 1998), reproductive effort (Stelzer, 2001), and offspring investment in relation to temperature (Stelzer, 2002). The B. calyciflorus clone was obtained from K.O. Rothhaupt and is identical to the B. calyciflorus used in his work (e.g. Rothhaupt, 1990a). Information on population growth and TFLs was established in an earlier publication (Stelzer, 1998). Rotifers were raised on Cryptomonas erosa var. reflexa Skuja, which was obtained from J.J. Gilbert, New Hampshire. In all experiments the rotifers were cultured in ADaM medium (Klüttgen et al., 1994), supplemented with 2.2 mg L )1 Na 2 EDTA and WC medium (Guillard, 1975) at a volume ratio of 9 : 1 to improve the conditions for Cryptomonas (Kirk & Gilbert, 1990). Cryptomonas erosa was cultured in WC medium in semicontinuous culture (dilution rate ¼ 0.25 day )1 ) and was continuously illuminated (100 lmol quanta m )2 s )1 ). The carbon content of C. erosa was 219 ± 8.7 pg cell )1 and the average cell volume was 885 lm 3 (equivalent spherical diameter approximately 12 lm). Populations of rotifers were acclimated to the experimental temperatures for at least 2 weeks in 2-L Erlenmeyer flasks and at a food concentration of 3 mg C L )1. Rotifers were cultured in continuous dim light (5 lmol quanta m )2 s )1 ) to minimise algal growth in the cultures. Feeding rates The functional response of both rotifer species was determined at 12 and 20 C. Rotifers were introduced into Cryptomonas suspensions at concentrations ranging from 0.1 to 1 mg C L )1, which were provided in tissue culture plates (volume: 8 ml). Feeding trials lasted for 12 h and at least five replicates were run for each combination of temperature and food concentration. The experimental animals originated from populations that were placed into a food concentration of mg C L )1 1 day prior to the experiment. In order to avoid additional grazing by newly hatched offspring, only large individuals without eggs were

4 2190 C.-P. Stelzer Table 1 Experimental design of the functional response experiments Food concentration [Cryptomonas cells ml )1 ] selected from these cultures. The exact age of these individuals was not known but most of them were probably young adults as they produced eggs during the feeding trial. The numbers of rotifers in the feeding trials ranged from four animals (Brachionus, 20 C, 0.1 mg L )1 ) to 185 animals (Synchaeta, 12 C, 1mgL )1 ). These numbers (see Table 1) had been estimated empirically from pilot experiments and were chosen to achieve a food depletion of approximately 10 50% during the incubation period. Control treatments without animals were run at all food concentrations and temperatures at the same time as the feeding trials. Algal concentrations were determined using an electronic particle counter (CASY, Schärfe GmbH, Reutlingen, Germany), set at a detection range of 8 15 lm equivalent spherical diameter. In an additional experiment, I determined the feeding rate for both rotifer species at a concentration of mg C L )1 (1500 cells ml )1 ), but at a wider range of temperatures (8, 10, 12, 16, 20 and 24 C). Filtration rates F were calculated according to: F ¼ mðlnc 0 lnc t Þ t where m is the water volume per individual, C 0 is the mean concentration of algae in the control treatments, C t is the concentration of algae at the end of the feeding trial, and t is the incubation time in days. The ingestion rate f was calculated according to the formula f ¼ F C m Brachionus Synchaeta 12 C 20 C 12 C 20 C Minimum and maximum number of individuals in the feeding trials (per 8 ml food suspension of each replicate). where C m is the average food concentration during the experiment. I estimated C m from initial and final algal densities in each replicate, assuming exponential decline during the experiment. Parameters of the functional response curves, such as the maximum feeding rate and the half-saturation constant, were estimated by fitting a curvilinear Michaelis Menten model to the untransformed data of ingestion rates versus algal concentrations. Starvation experiment Starvation resistance at different temperatures (12, 16, 20 and 24 C) was established for both rotifer species. Experiments were initiated with cohorts (hatching interval: h, depending on temperature) of 24 young adults that had been cultured at a food concentration of 1500 cells ml )1 until they reached maturity (2 7 days, depending on temperature). Animals were then transferred into food-free medium and survival was recorded in 12 h-intervals until the last animal had died. Controls with fed animals were run parallel to these experiments. Almost all individuals in these controls survived until the end, indicating that the mortality in the food-free treatments was caused by starvation rather than old age. In Brachionus, only the temperatures C were tested, as the population that was acclimated to 24 C inexplicably went extinct shortly before the experiment. Survival curves of both species were compared for each temperature separately using Gehan s Wilcoxon test for pairwise comparisons (Software: Statistica Ò, StatSoft Inc.). Simulation model Data from the functional response curves as well as data from the numerical response were used to parameterise a mathematical model of resource competition consisting of three differential equations representing the two consumers and their common resource: dn 1 N 1 dt ¼ r max1ðs S 01 Þ ðk s1 þ S S 01 Þ D pred dn 2 N 2 dt ¼ r max2ðs S 02 Þ ðk s2 þ S S 02 Þ D pred ds dt ¼ D preyðs v SÞ N 1 f max1 S K F1 þ S N 2 f max2 S K F2 þ S where N are the concentrations of the rotifers (Brachionus, Synchaeta), S is the concentration of the resource (Cryptomonas), r max is the maximum growth rate of the rotifers, K s is the half-saturation constant of the numerical response, S 0 is the TFL for zero population

5 Temperature and competition of rotifers 2191 growth, D pred is the dilution rate applied to the rotifers, f max is the maximum ingestion rate of the rotifers, K F is the half-saturation constant of the functional response, D prey is the dilution rate applied to the algae, and S v is the concentration of the algae in the food supply. Note that algal concentrations in this model can only increase because of inflow from the food supply and that intrinsic algal growth is neglected. The model was parameterised using estimates of feeding rates from this study and estimates of population growth from an earlier study (Stelzer, 1998). The various parameter estimates are displayed in Table 2. Numerical simulations were carried out using initial densities of 1 individual ml )1 for each species and for an algal concentration of mg CL )1 (1500 cells ml )1 ) as food supply (S v ). The model was implemented in Matlab Ò (The MathWorks Inc.) and iterative calculations were run in time steps of 0.1 day. Competition experiments Two sets of competition experiments were run to test the predictions about competitive dominance between S. pectinata and B. calyciflorus at the different temperatures. In the first experiment, the animal populations were cultured in six-well tissue culture plates (8 ml volume). The Cryptomonas concentration in the supplied medium was mg C L )1 (1500 cells ml )1 ). Initial densities of the animals were 1 individual ml )1. The experiment was initiated with a mixture of juveniles and adults (ratio 50 : 50) taken from a culture with a food concentration of mg C L )1. At intervals of 48 h all rotifers and their eggs were counted and transferred into a new vessel with fresh food suspension. The experiment was run at two temperatures (12 and 20 C) and two competition regimes (alone, with competitor). Three replicates were performed of each treatment. The main difference in the second competition experiment was the time interval at which the Cryptomonas solution was supplied. In this case, one-third of the culture medium was replaced by fresh food suspension every 8 h, resulting in a closer approximation to a continuous food supply. Rotifers were cultured in glass bottles (1100 ml volume) placed on plankton wheels, rotating every 15 min at 1 rpm for 2 min. The inoculation of each of these cultures was carried out with a mixture of 1100 juveniles and adults, in a similar manner to the first competition experiment. At intervals of 8 h, one-third of the medium was replaced by fresh food suspension (1500 cells ml )1 ) by filtration through 30-lm gauze. Abundance of the two rotifer species and egg ratios were determined from daily samples of 50 ml volume. Egg ratio was defined as the number of eggs divided by the number of females in a sample. These were the only occasions when rotifers were harvested, resulting in a dilution rate of approximately 0.05 day )1. This experiment was run only at 12 C, as, given the strong competitive dominance of Brachionus in the first experiment, the question shifted towards whether Synchaeta could outcompete Brachionus at the lower temperature. Furthermore, for logistic reasons parallel trials at 20 C were not possible. The experiment had three treatments (Synchaeta alone, Brachionus alone, both species together) and three replicates per treatment. Residual algae Table 2 Parameters used in the competition model 12 C 20 C Synchaeta Brachionus Synchaeta Brachionus Numerical Response r max [day )1 ] K s [mg C L )1 ] S 0 [mg C L )1 ] Functional response f max [lg C individual )1 day )1 ] K F [mg L )1 ] Data based on Stelzer (1998) and the ingestion rates measured in this paper (see Fig. 3). r max, maximum growth rate; K s, half-saturation constant of the numerical response; S 0, threshold food level for population growth; f max, maximum feeding rate; K F, half-saturation constant of the functional response.

6 2192 C.-P. Stelzer (i.e. the algal concentrations after an 8-h interval) were measured once per day using the electronic particle counter. This measurement provided an estimate of the minimal algal concentration each day. Maximal algal concentrations were calculated by taking into account the fresh food suspension that was replaced every 8 h. Statistical analysis of the competition experiments was carried out for each temperature separately, using 2-way ANOVAs with time as a repeated-measures factor to test if the presence of the competitor had a significant effect on population density of the focal species. In the second competition experiment, egg ratios of S. pectinata showed conspicuous fluctuations throughout the experiment. To test if these fluctuations had a periodic component, autocorrelation functions (ACFs) were calculated for each replicate (Software: Statistica Ò, time series analysis module). Prior to the autocorrelation analysis, data sets were lntransformed and linear trends were removed, using the software s standard functions. Cycles in egg ratios were considered statistically significant if the ACF oscillated between positive and negative values and exceeded the 95% confidence limits at a time lag equal to the period length (Turchin & Taylor, 1992). Results Feeding rates of B. calyciflorus increased strongly with temperature, whereas those of S. pectinata stayed relatively constant at temperatures of 8 24 C (Fig. 2). For instance, from 8 to 20 C the ingestion rates of Brachionus at a food concentration of mg C L )1 increased 6.2-fold, while in Synchaeta the increase was only 1.7-fold. This difference is reflected in a significant interaction of the factors species temperature in the two-way ANOVA (F 6,55 ¼ 37.46, P < 0.01). Across the temperature range studied, feeding rates of Brachionus were generally higher than those of Synchaeta (up to 4.6-fold higher at 20 C), although the difference between the two species diminished markedly at the lower temperatures. The higher feeding rates of Brachionus were also apparent under a wider range of food concentrations (see functional response curves, Fig. 3). Starvation time in both rotifer species decreased with increasing temperature (Fig. 4). There was a strong difference in starvation resistance between the two rotifer species. At 12 C, starvation times of Fig. 2 Temperature-dependence of the ingestion rates of Synchaeta and Brachionus at a food concentration of mg C L )1 (1500 cells ml )1 Cryptomonas). Means and standard errors. Fig. 3 Functional response curves of Synchaeta and Brachionus at 12 C and 20 C. Symbols: measured values; fitted lines: nonlinear regression according to Michaelis Menten models. Fig. 4 Temperature-dependence of starvation times of adult Synchaeta and Brachionus. Mean and standard deviations. Brachionus were almost 2.3 times longer than those of Synchaeta. At 20 C this difference was weaker (1.5- fold). Mean survival at 20 C was 2.3 days in Brachionus and 1.5 days in Synchaeta. The differences between species were statistically significant at all three temperatures, from 12 to 20 C (Gehan s Wilcoxon test for pairwise comparisons, P < 0.001).

7 Temperature and competition of rotifers 2193 Fig. 5 Theoretical predictions of the time course of competition between Synchaeta and Brachionus at 12 and 20 C. The competition model is based on the numerical response curves and functional response curves of the two competitors (parameter values, see Table 2). The competition model was parameterised with the functional response data presented in Fig. 3 and data on growth rates of Brachionus and Synchaeta from an earlier publication (Stelzer, 1998). A summary of these parameter values is presented in Table 2. Initial densities of the competitors were set to 1 individual ml )1 and food concentration in the supply S v was set to mg C L )1. Dilution rates were set to D prey ¼ 1 day )1 for the algae and D pred ¼ 0.05 day )1 for the rotifers, which mimics the experimental conditions in the second competition experiment. In the simulations Synchaeta was competitively displaced by Brachionus within 3 days at 20 C (Fig. 5). In contrast, at 12 C the simulation predicted Synchaeta to be the superior competitor. However, in this case the predicted time to competitive exclusion was much longer (30 days), although the Brachionus population was predicted to decline steadily after the fifth day. In the first competition experiment Brachionus was the superior competitor at both temperatures (Fig. 6). Synchaeta was excluded after 4 days at 20 C and after 16 days at 12 C. In the controls without Brachionus, Synchaeta grew exponentially and built up large populations, up to 8 individuals ml )1 at 20 C and 12 individuals ml )1 at 12 C. The Brachionus populations grew more slowly in the presence of Synchaeta but this trend completely disappeared as soon as the Synchaeta populations died out. Afterwards the Brachionus populations in the competition treatment were not distinguishable from those in the single species treatments. Measurements of residual algae after each transfer interval showed that in the competition treatments algal concentrations fell below the detection limits of the electronic particle counter after only 2 days at 20 C and after 6 days at 12 C (data not shown). The second competition experiment was run at 12 C and was aimed to test the hypothesis that Synchaeta could outcompete Brachionus if they are provided with a feeding regime approximating a continuous food input (every 8 h). In all three replicates both competitors coexisted for 32 days and there Fig. 6 Results of the first competition experiment. Time courses of abundances of Synchaeta and Brachionus in populations with and without competitors. Mean and standard deviations.

8 2194 C.-P. Stelzer Fig. 7 Results of the second competition experiment. Dynamics of the Brachionus (left) and Synchaeta populations (right). Top: egg ratios; middle: rotifer abundances; bottom: maximal and minimal food concentrations in the competition treatments (lines), threshold food levels of the two rotifer species (dashed lines). was no indication of competitive exclusion in either species (Fig. 7). However, both rotifer species were negatively affected by the presence of their competitor, as they build up smaller populations in mixed species cultures versus single species cultures (see Fig. 7; Table 3). During the 6 days of the experiment, when population size remained relatively stable in both species, Synchaeta populations were reduced on average by 70% in the presence of Brachionus, while Brachionus populations were reduced on average by 43% in the presence of Synchaeta. Concentrations of the food algae Cryptomonas fluctuated during the experiment, but these fluctuations took place over a relatively small range as can be seen from the maximal and minimal food concentrations in Fig. 7 (bottom graph). Cryptomonas concentrations declined steadily until they approached the TFL of Syncheta. After day 15, they were permanently below the TFL of Brachionus. Interestingly the Brachionus population was not decreasing at this time. In fact, it slightly increased in the second half of the experiment. In the control populations without Synchaeta, Brachionus also continued to grow after the Cryptomonas concentrations had already fallen below the TFL (Fig. 8). In contrast, the control populations of Synchaeta showed the expected pattern of population stagnation as Cryptomonas concentrations approached the TFL (Fig. 8). Dependent variable Factor d.f. F-value P-value Brachionus [individual ml )1 ] Syn 1, <0.01 Time 31, <0.001 Syn time 31, <0.001 Synchaeta [individual ml )1 ] Brach 1, <0.001 Time 31, <0.001 Brach time 31, <0.001 Table 3 Results of the repeated-measures ANOVAs on rotifer abundances in the second competition experiment

9 Temperature and competition of rotifers 2195 Fig. 8 Second competition experiment: dynamics in the single treatments (without competitors). Population dynamics of Brachionus and Synchaeta and concentrations of Cryptomonas erosa. Mean abundances (circles) and maximal/minimal food concentrations (dotted lines). Egg ratios showed large fluctuations as well as an overall decline during the experiment (Fig. 7). Especially in Synchaeta these fluctuations in egg ratios seemed to have a periodic component (see Fig. 7, top right). Autocorrelation analysis showed that these cycles were significant in at least four replicates (data not shown), with the period of one cycle being 6 9 days. In the two remaining replicates the same trend was apparent but it was not statistically significant. In Brachionus there were no strong hints of a cyclic pattern in egg ratios. Only one replicate exhibited significant cycles of the ACF, with the remaining replicates showing no trends for periodicity. Discussion The main goal of this paper was to test the hypothesis that competitive abilities and the outcome of resource competition can reverse along a temperature gradient. Previous studies suggested that this should be the case when the two rotifers S. pectinata and B. calyciflorus compete for the food algae C. erosa at different temperatures (Stelzer, 1998). At low temperatures (12 C) the TFL of Synchaeta was significantly lower than that of Brachionus and this situation was reversed at 20 C. According to resource competition theory Synchaeta should displace Brachionus at 12 C and Brachionus should displace Synchaeta at 20 C. The results of this paper did not confirm this hypothesis. In the first competition experiment Brachionus displaced Synchaeta at both 12 and 20 C. And in the second experiment, which was conducted at 12 C, Synchaeta did not displace Brachionus. The species coexisted for more than 1 month and, although the abundances of both species were negatively affected by the presence of their competitor, there was no indication that Synchaeta would become dominant. The weak competitive performance of Synchaeta at 12 C may require different explanations in the two competition experiments and will be discussed separately below. In the first competition experiment, resources were supplied in a discontinuous manner. Every second day all animals were transferred into a new vessel containing food suspension at a concentration of mg C L )1. This concentration is well above the TFL of both species, but because of population growth and grazing it was reduced within a short time after the transfers. The discontinuous food supply probably favoured Brachionus for two reasons. First, Brachionus had higher feeding rates than Synchaeta and could therefore ingest more food within a short time. Second, Brachionus had a higher starvation tolerance and was better able to persist through the time periods of extremely low food abundance, which preceded each transfer interval in the later phase of the experiment. The reason for the difference in starvation tolerance probably lies in the contrasting reproductive strategies of the two species in response to food deprivation: Brachionus reduces its reproductive output and lowers its metabolism to survive periods of low food while Synchaeta actually increases reproduction at the expense of adult survival (Kirk, 1997; Stelzer, 2001, 2005). Fluctuating resource abundance can lead to competitive outcomes that deviate greatly from those predicted by the TFLs if the competitors show differential resistance to starvation (MacIsaac & Gilbert, 1991; Kirk, 2002). These mechanisms may have been responsible for the strong dominance of Brachionus in the first competition experiment.

10 2196 C.-P. Stelzer Interestingly, many studies of resource competition have used an experimental protocol that avoids complications arising from starvation conditions. In such studies relatively high dilution rates of day )1 were applied to the rotifer populations (Rothhaupt, 1988; Ciros-Perez et al., 2001). An argument often made is that this mimics the natural mortality imposed by predators. However, as a result, the food concentrations that are necessary to maintain population equilibrium are much higher than the TFL. Accordingly, the inferior competitors are in a physiological state that cannot be termed starvation, even at times when they are being competitively displaced. Rather, they are excluded from the system because they do not grow fast enough to avoid the washout because of dilution. In contrast, the dilution rates of the rotifers in this study were much lower ( day )1 ), so the animals were closer to their actual TFL. Thus their population dynamics may have been determined mainly by their physiological response to starvation. This suggests that estimates of population growth rate (such as those obtained from a curve fit using the Monod-model) may not be accurate enough to predict long-term population dynamics in situations where resource concentrations are close to the TFL. In the second competition experiment C. erosa was supplied at very short time intervals (8 h), which closely mimicked a continuous supply of food. Thus, resource fluctuations and differential tolerance to starvation cannot serve as explanations in this experiment. Measurements of the residual Cryptomonas concentrations indicated that resource concentrations fluctuated between feeding bouts, but within limits that were smaller than the difference between the TFLs of the two rotifer species. This means that from the middle of the experiment (approximately day 15) food concentrations were permanently below the TFL of Brachionus. However, the Brachionus population did not decrease as predicted from resource competition theory. In both the single species treatment and in the competition treatment, Brachionus populations were stable even when the concentrations of Cryptomonas were far below the TFL. In contrast Synchaeta showed exactly the expected dynamics: populations grew exponentially until the Cryptomonas concentrations approached its TFL and then reached a stable equilibrium (Fig. 8). A possible explanation for the unusual behaviour of Brachionus is that they also fed on detritus and associated bacteria, which were inevitably accumulating during the later phase of the experiment. Brachionus is a generalistic consumer, feeding on algae of a wide range of sizes and it can even grow on bacteria (Starkweather, Gilbert & Frost, 1979). In contrast, Synchaeta is more specialised and can not ingest bacteria (Gilbert & Bogdan, 1984). Brachionus may thus have supplemented its diet with this additional food source. Such effects were probably absent in the experiments where the numerical response of Brachionus was established, as the culture medium was exchanged completely each day (Stelzer, 1998). Thus, in the second competition experiment there may have been resource partitioning that was not taken into account by the competition model. Several studies have demonstrated that resource partitioning can facilitate coexistence among consumers (Rothhaupt, 1988; Ciros-Perez et al., 2001). A methodological improvement of the second competition experiment would be monoxenic cultures, involving only the rotifers and their food algae Cryptomonas, but without bacterial contaminants. Such monoxenic cultures have been described for rotifers (e.g. Gilbert, 1970), but are extremely difficult to maintain over long periods. It is theoretically possible that the TFL of Brachionus at 12 C was overestimated in the earlier study (Stelzer, 1998) and that this explains the good performance and the below-tfl population growth. Comparisons with literature data are restricted to 20 C; however these do not indicate that the TFLs were overestimated. In fact, the value for the TFL of Brachionus at 20 C of 0.13 mg C L )1 was even lower than that estimated by Rothhaupt (1990a) for a similarly sized algae, C. sphaeroides (TFL: 0.19 mg CL )1 ). Stemberger & Gilbert (1985) obtained a similar value for B. calyciflorus at 19 C (TFL: 0.38 mg L )1 dry weight of C. erosa). Also the TFL estimate for Synchaeta at 20 C of 0.28 mg C L )1 comes very close to the 0.5 mg dry weight L )1 obtained by Stemberger & Gilbert (1985) assuming approximately 50% carbon content in the dried algal biomass. Thus, it seems rather unlikely that the unexpectedly good performance of Brachionus at low food concentrations was because of an overestimation of its TFL. Data on growth and feeding suggest that S. pectinata, or at least the clone studied, is physiologically adapted to low temperatures. The Synchaeta clone can maintain positive population growth rates at temperatures as low as 4 C (Stelzer, 2002). Data on growth rates of the Brachionus clone at such low temperatures are not

11 Temperature and competition of rotifers 2197 available, but Halbach (1970) estimated 10 C as the lowest temperature for population development in B. calyciflorus. If population growth rates are compared for different temperatures, Synchaeta reached 65% of its maximum growth rate at 12 C, whereas Brachionus reached only 24% at the same temperature (Stelzer, 1998). Similarly, the relationship of ingestion rates versus temperature indicates that Synchaeta was feeding at close to maximum rates at 8 C whereas the feeding rates of Brachionus showed a strong increase from 8 to 20 C (Fig. 2). Similar indications of temperature adaptation were found by Gilbert & Bogdan (1984), who measured filtration rates of various rotifers in situ. They found monotonically increasing filtration rates with temperature in Keratella cochlearis Gosse and two Polyarthra species. Additionally they showed that the filtration rates of the cold-stenothermal Polyarthra dolichoptera Idelson at 7 C were already as high as those of the eurythermal Polyarthra vulgaris Carlin at 20 C (Gilbert & Bogdan, 1984). The relationship between TFL and temperature may be an indicator of the temperatures to which an organism is adapted. In resource limited environments there is selection for very efficient use of food. Hence in a species that often faces resource limitation, the zones with the lowest TFLs may be a better indicator of the optimal temperature range than reaction norms that display growth rate versus temperature at high food concentrations. Several studies have demonstrated that rotifers, especially S. pectinata, are often food limited in their natural environment (Gonzalez & Frost, 1992; Merriman & Kirk, 2000; Stelzer, 2001). TFLs are expected to increase at higher temperatures if the assimilation rate has a temperature optimum but the rate of metabolic losses increase steadily with temperature (Lampert, 1977). The strong increase of TFL in Synchaeta may be caused by such mechanisms (Fig. 1). Over a very wide range of temperatures, optimum curves (U-shaped curves) are expected, in which the TFL stays constantly low over a preferred range of temperatures and only increases at the extremes. Such reaction norms have been described in cladocerans (Achenbach & Lampert, 1997) and a ciliate (Weisse et al., 2002). An unexpected observation during the second competition experiment was the occurrence of periodic oscillations in the egg ratios of Synchaeta. Itis unlikely that the egg ratio oscillations were triggered by fluctuations in resource abundance. First, resource concentrations (estimated as residual algae in the experimental vessels) did not show any cyclic fluctuations at all, but declined monotonically throughout the experiment. Second, the tendency for oscillations in egg ratios seemed to be much more pronounced in Synchaeta than in Brachionus, although both rotifers shared the same resource environment in the replicates of the competition treatment. The egg ratio oscillations were probably caused by the specific fecundity pattern of Synchaeta. Life table analysis of S. pectinata shows that production of offspring is maximal in the first adult age classes and then gradually decreases, displaying a triangular fecundity pattern (Stelzer, 2001). In contrast, offspring production in Brachionus seems to be constantly high across the adult age classes, forming a rectangular fecundity pattern (Halbach, 1970). This difference in fecundity schedule may be crucial, as a concentration of reproduction in a few age classes can slow down the convergence to a stable age structure (Caswell, 1989). A theoretical example of such cyclic age structure fluctuations is displayed in Fig. 9. Two hypothetical organisms are assumed whose life cycles are represented by the two Leslie matrices in Fig. 9. In the first organism, reproduction is concentrated in the first adult age class and then gradually declines (triangular fecundity pattern) whereas in the second organism reproduction is constant across all adult age classes (rectangular fecundity pattern). Both organisms have identical survival schedules and the same overall population growth rate (r ¼ 0.092). The projections of their Leslie matrices reveals that both organisms exhibit dampened oscillations in age structure, represented by the ratio between the youngest age class and the sum of all remaining age classes (¼egg ratio). In the organism with the triangular fecundity pattern these oscillations are more persistent than in the organism with the rectangular fecundity pattern, which shows one cycle with large amplitude at the beginning and then quickly converges to a stable egg ratio. Similar mechanisms may have caused the differential tendency of Synchaeta and Brachionus to form periodic cycles in egg ratios (Fig. 7). In conclusion, the results of this study suggest that several mechanisms may cause deviations in the outcome of competition, as opposed to that predicted by the TFL. Resource fluctuations and differential

12 2198 C.-P. Stelzer Fig. 9 Leslie matrices and periodic cycles of the egg ratios in two hypothetical organisms. Top: Organism with triangular fecundity pattern, showing fecundities of 1, 0.4 and 0.1 in the adult age classes. Bottom: Organism with rectangular fecundity pattern, showing fecundities of 0.66 in all adult age classes. The projections on the right show the temporal dynamics of the egg ratio for each Leslie matrix. Egg ratio is defined as the ratio between the abundance of the first age class divided by the sum of abundances in all remaining age classes. Starting condition: five individuals per age class. tolerance of starvation may even reverse the outcome of competition. Unaccounted resource partitioning, such as in the second experiment, may cause additional deviations from theoretical predictions. It may be argued that such deviations could have been avoided by a finer tuning of the experimental conditions. However, resource fluctuations and complex food situations actually occur in most natural systems and are thus of ecological relevance. Finally, demographic effects, such as egg ratio oscillations represent another deviation from an assumption of resource competition theory, where populations are assumed to be in their stable age distribution. It remains to be investigated whether such deviations can actually affect the outcome of resource competition. Interestingly there is evidence from Daphnia pulex showing that temporal fluctuations in age structure can have strong impact on population dynamics and competition among clones (Nelson, McCauley & Wrona, 2005). Acknowledgments I thank W. Lampert and M. Boersma for comments on an earlier version of the manuscript. Financial support was provided by DFG grant LA 309/12 1. References Achenbach L. & Lampert W. (1997) Effects of elevated temperatures on threshold food concentrations and possible competitive abilities of differentially sized cladoceran species. Oikos, 79, Berzinš B. & Pejler B. (1989) Rotifer occurrence in relation to temperature. Hydrobiologia, 175, Caswell H. (1989) Matrix Population Models. Sinauer, Sunderland. Ciros-Perez J., Carmona M.J. & Serra M. (2001) Resource competition between sympatric sibling rotifer species. Limnology and Oceanography, 46, Davies A.J., Lawton J.H., Shorrocks B. & Jenkinson L.S. (1998b) Individalistic species responses invalidate simple physiological models of community dynamics under global environmental change. Journal of Animal Ecology, 67, Davies A.J. Jenkinson L.S., Lawton J.H., Shorrocks B. & Wood S. (1998a) Making mistakes when predicting shifts in species range in response to global warming. Nature, 391, Gilbert J.J. (1970) Monoxenic cultivation of the rotifer Brachionus calyciflorus in a defined medium. Oecologia, 4, Gilbert J.J. & Bogdan K.G. (1984) Rotifer grazing: In situ studies on selectivity and rates. In: Trophic Interactions within Aquatic Ecosystems (Eds D.G. Meyers & J.R. Strickler), pp American Association for the Advancement of Science, Selected Symposium 85, Boulder, Colorado, U.S.A. Gonzalez M.J. & Frost T.M. (1992) Food limitation and seasonal population declines of rotifers. Oecologia, 89,

13 Temperature and competition of rotifers 2199 Guillard R.R.L. (1975) Culture of phytoplankton for feeding marine invertebrates. In: Culture of Marine Invertebrate Animals (Eds W.L. Smith & M.H. Chanley), pp Pleum Pub. Co., New York. Halbach U. (1970) Einfluß der Temperatur auf die Populationsdynamik des planktischen Rädertieres Brachionus calyciflorus. Oecologia, 4, Hughes L. (2000) Biological consequences of global warming: is the signal already apparent? Trends in Ecology & Evolution, 15, Jiang L. & Morin P.J. (2004) Temperature-dependent interactions explain unexpected responses to environmental warming in communities of competitors. Journal of Animal Ecology, 73, Kirk K.L. (1997) Life-history responses to variable environments: starvation and reproduction in planktonic rotifers. Ecology, 78, Kirk K.L. (2002) Competition in variable environments: experiments with planktonic rotifers. Freshwater Biology, 47, Kirk K. & Gilbert J.J. (1990) Suspended clay and the population dynamics of planktonic rotifers and cladocerans. Ecology, 71, Klüttgen B., Dülmer U., Engels M. & Ratte H.T. (1994) Corrigendum. Water Research, 28, 1. Kreutzer C. & Lampert W. (1999) Exploitative competition in differentially sized Daphnia species: a mechanistic explanation. Ecology, 80, Lampert W. (1977) Studies on the carbon balance of Daphnia pulex de Geer as related to environmental conditions. IV. Determination of the threshold concentration as a factor controlling the abundance of zooplankton species. Archiv für Hydrobiologie Supplement, 48, MacIsaac H.J. & Gilbert J.J. (1991) Competition between Keratella cochlearis and Daphnia ambigua: effects of temporal patterns of food supply. Freshwater Biology, 25, Merriman J.L. & Kirk K.L. (2000) Temporal patterns of resource limitation in natural populations of rotifers. Ecology, 81, Nelson W.A., McCauley E. & Wrona F.J. (2005) Stagestructured cycles promote genetic diversity in a predator-prey system of Daphnia and algae. Nature, 433, Rothhaupt K.O. (1988) Mechanistic resource competition theory applied to laboratory experiments with zooplankton. Nature, 333, Rothhaupt K.O. (1990a) Population growth rates of two closely related rotifer species: effects of food quality, particle size, and nutritional quality. Freshwater Biology, 23, Rothhaupt K.O. (1990b) Differences in particle size-dependent feeding efficiencies of closely related rotifer species. Limnology and Oceanography, 35, Starkweather P.L., Gilbert J.J. & Frost T.M. (1979) Bacterial feeding by the rotifer Brachionus calyciflorus: clearance and ingestion rates, behavior and population dynamics. Oecologia, 44, Stelzer C.P. (1998) Population growth in planktonic rotifers: does temperature shift the competitive advantage for different species? Hydrobiologia, 387/ 388, Stelzer C.P. (2001) Resource limitation and reproductive effort in a planktonic rotifer. Ecology, 82, Stelzer C.P. (2002) Phenotypic plasticity of body size at different temperatures in a planktonic rotifer: mechanisms and adaptive significance. Functional Ecology, 16, Stelzer C.P. (2005) Evolution of rotifer life histories. Hydrobiologia, 546, Stemberger R.S. & Gilbert J.J. (1985) Body size, food concentration, and population growth in planktonic rotifers. Ecology, 66, Stenseth N.C., Mysterud A., Ottersen G., Hurrell J.W., Chan K.S. & Lima M. (2002) Ecological effects of climate fluctuations. Science, 297, Tilman D. (1982) Resource Competition and Community Structure. Princeton University Press, Princeton, New Jersey, U.S.A. Tilman D., Mattson M. & Langer S. (1981) Competition and nutrient kinetics along a temperature gradient: an experimental test of a mechanistic approach to niche theory. Limnology and Oceanography, 26, Turchin P. & Taylor A.D. (1992) Complex dynamics in ecological time series. Ecology, 73, Weisse T., Stadler P., Lindström E.S., Kimmance S.A. & Montagnes D.J.S. (2002) Interactive effect of temperature and food concentration on growth rate: a test case using the small freshwater ciliate Urotricha farcta. Limnology and Oceanography, 47, (Manuscript accepted 1 August 2006)